- Email: [email protected]
Incidence of head and neck cancer and socioeconomic status in Canada from 1992 to 2007
Oral Oncology 49 (2013) 1072–1076
Contents lists available at ScienceDirect
Oral Oncology journal homepage: www.elsevier.com/locate/oraloncology
Incidence of head and neck cancer and socioeconomic status in Canada from 1992 to 2007 Euna Hwang a, Stephanie Johnson-Obaseki a, James Ted McDonald b, Chris Connell b, Martin Corsten a,⇑ a b
Department of Otolaryngology-Head and Neck Surgery, University of Ottawa, Ottawa, Canada Department of Economics, University of New Brunswick, Fredericton, Canada
a r t i c l e
i n f o
Article history: Received 25 January 2013 Received in revised form 5 May 2013 Accepted 14 August 2013 Available online 7 September 2013 Keywords: Head and neck cancer Incidence Socioeconomic status Income Oropharynx cancer Human papillomavirus
s u m m a r y Objective: To investigate the association between head and neck cancer (HNC) incidence and socioeconomic status (SES) in Canada, and investigate changes in the relationship between HNC incidence and SES from 1992 to 2007. Materials and methods: Cases were drawn from the Canadian Cancer Registry (1992–2007), and were categorized into three subsites: oropharynx, oral cavity, and ‘‘other’’ (hypopharynx, larynx, and nasopharynx). Demographic and socioeconomic information were extracted from the Canadian Census of Population data for the study period (1992–2007), which included four census years. We linked cases to income quintiles (InQs) according to patients’ postal codes. Results: Incidence rates in the lowest InQs were significantly higher than in the highest InQs for all HNC subsites. The incidence of oropharyngeal cancer increased over the time period 1992–2007 for all InQs. However, the greatest increase in incidence was in the highest InQs. As a result, over the time period the gap between the incidence of the highest and lowest InQs significantly narrowed for oropharyngeal cancer. For oral cavity cancer and the other head and neck cancers, the overall incidence did not increase and the gap in incidence did not change significantly. Conclusion: HNC incidence was higher with lower income in all HNC subsites in Canada from 1992 to 2007. The gap in incidence between the highest and the lowest InQs narrowed for oropharyngeal cancer over the time period studied, but was unchanged for the other HNC subsites. Ó 2013 Elsevier Ltd. All rights reserved.
Introduction Head and neck cancer (HNC) accounts for more than 550,000 cases annually worldwide and represents approximately 3% of malignancies in the United States (US) [1,2]. Multiple studies from the US, Canada, and Europe have revealed that HNC is more frequent in socioeconomically-deprived populations [4–9]. Although this observation is mostly attributed to confounding risk factors associated with lower socioeconomic status (SES) such as tobacco and alcohol consumption and poor oral hygiene, our group and others have shown that there is a higher incidence of HNC with lower SES after controlling for such factors [4,7,8]. The underlying etiology behind this association remains unclear and is likely multifactorial. Moreover, there is no consensus in the literature for parameters of SES when studying the relationship between SES and the epidemiology of cancer.
⇑ Corresponding author. Address: Department of Otolaryngology-Head and Neck Surgery, University of Ottawa/The Ottawa Hospital, Suite S-3, 501 Smyth Road, Ottawa, ON K1H 8L6, Canada. Tel.: +1 (613) 737 8596; fax: +1 (613) 739 6542. E-mail address: [email protected] (M. Corsten). 1368-8375/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.oraloncology.2013.08.002
Our group previously examined the association between SES and HNC incidence in both the US (National Institutes of Health Survey) and Canada (case-control study conducted in the province of Ontario) by looking at various measures of SES. In the US, the following factors were associated with a higher incidence of HNC: marital status of single, never married; lower education (maximum level of education achieved less than high school completion); as well as lower income (family income of less than US$20,000) [4]. In Ontario, Canada, an increased incidence of HNC was similarly associated with lower education (less than grade 8 education) and lower median family income [8]. When specifically assessing income, MacKillop et al. concluded that there were strong, inverse associations between income level and the incidence of cancers of the head and neck region in Ontario and the US [3]. In addition to the effects of SES on HNC incidence, there have been dramatic changes in the incidence of cancer at some head and neck subsites, particularly oropharynx, over the last two decades. This change in incidence is related to the increase in the number of head and neck cancers caused by human papilloma virus over the same time period. On the other hand, the incidence of cancer at other head and neck subsites, such as larynx, has been
E. Hwang et al. / Oral Oncology 49 (2013) 1072–1076
decreasing over that time period, largely due to the decrease in smoking incidence over that time. The incidence of oral cavity cancer has been relatively stable. Our group sought to investigate the in the SES/HNC incidence relationship over the time period 1992– 2007. Our a priori hypothesis was that the magnitude of the relationship between SES and HNC incidence for oropharynx cancer would decrease over the time period, whereas the magnitude of the relationship between SES and HNC incidence for other head and neck cancers not predominantly caused by HPV would remain stable. To our knowledge, there has not been a Canadian study using nationwide Cancer Registry data to examine the relationship between HNC incidence and SES. The goals of the study were twofold: (1) to confirm the relationship between SES and HNC incidence using Canadian national Registry data and; (2) to investigate the changes in the relationship between SES and HNC incidence over time.
Patients and methods Our data were drawn from two sources: the Canadian Cancer Registry (CCR) data file and the Canadian Census of Population from Statistics Canada. The CCR data file contains patient demographic and tumor-specific data on each tumor recorded in provincial and territorial cancer registries from 1992 to 2007, inclusive. Names and personal identification numbers were removed and replaced with unique identifiers in the version of the file released for this study. All analysis was conducted within the New Brunswick Research Data Centre (NB-RDC) and all output was vetted for release using enhanced vetting methods required by Statistics Canada. Our study complied with the University of New Brunswick Institutional Review Board ethics requirements, which do not necessitate an additional review for research projects using Statistics Canada data stored in the NB-RDC. To our knowledge, this is the first time CCR data have been available and used to study the relationship between HNC incidence and SES in Canada. The Canadian Census of Population contains demographic and socioeconomic information from each census, which occurs every five years in Canada. Data were available for the census years (CYs) 1991, 1996, 2001, and 2006. The finest level of disaggregation at which census information is released by Statistics Canada is the Dissemination Area (DA), a small, stable geographic unit composed of one or more adjacent blocks, with a population of 400–700 persons. DAs cover the whole territory of Canada. For each DA and for each of the four CYs, information was collected on age for men and women, average household income, and proportion of adult residents with at least a university degree. Although average income in the DA might be influenced by large income outliers, median income is not available at the DA-level for all of the CYs. The number of DAs for which census information was available was 32,825 in 1991, 38,016 in 1996, 46,909 in 2001, and 52,443 in 2006. In order to control for wage and price differences across different regions, the income quintile (InQ) for each DA was defined relative to other DAs in the associated Census Division (CD), which Statistics Canada defines as a group of neighboring municipalities joined together for the purposes of regional planning and managing common services. The number of CDs in Canada was in the range of 288–290 across the four CYs and changes occurred to only a few CD boundaries each CY. DAs within each CD were sorted by average income and then assigned to one of five InQs. DAs in a CD are ranked by average household income from highest to lowest. Each DA is assigned to an InQ based on its rank in the distribution. Measuring with respect to the CD allows for variation in prices and purchasing power across Canadian regions. Although the DA of an individual in the CCR database
1073
is not reported, the postal code of residence is disclosed. PCCF+ is a statistical tool provided by Statistics Canada that maps postal codes to DAs. Statistics Canada revises geographic boundaries and updates PCCF+ at each CY. Hence, in order to match postal code and DA information as closely as possible, HNC cases were assigned to CYs as follows: cases diagnosed in 1992–1995 were associated with socioeconomic characteristics according to the 1991 Census; cases diagnosed in 1996–2000 were associated with data from the 1996 Census; cases diagnosed in 2001–2005 were associated with data from the 2001 Census; and cases diagnosed in 2006–2007 were associated with data from the 2006 Census. Census data could be linked to 96.9% of DAs for the CY 1991, 97.4% for the CY 1996, 99.0% for the CY 2001, and 99.2% for the CY 2006. For each CY, the unit of observation for the analysis of incidence was the DA. The key variable of interest was the number of cases of different types of HNC diagnosed in adults over the age of 18 in each DA over the relevant period of time corresponding to that CY. At the DA-level, it is not feasible to disaggregate counts further by age and gender. In the regression analysis, controls were included for the age and gender composition of the adult population of each DA. The exposure variable was the adult population in the DA at the CY multiplied by the number of years in the corresponding time interval for that census (2, 4, or 5 years). Poisson regression models were estimated where neighborhood SES as measured by the InQ of the DA was captured by a 0/1 binary variable for each InQ, with the highest InQ specified as the baseline. The regressions also included indicator variables for each province or territory of residence in order to capture regional unobserved effects important to the incidence and diagnosis of HNC. To capture changes in incidence over time, indicator variables were defined for each CY from 1996 to 2006 with 1991 as the reference year. These CY variables interacted with each InQ to allow the incidence of HNC to vary by average household income over time. One of the main risk factors for most forms of HNC is smoking; however, data on smoking prevalence was not available at the DAlevel from Statistics Canada surveys. As a substitute, provincial smoking rates by education level were obtained for each CY from various waves of the Canadian Community Health Survey and the National Population Health Survey. For each DA in our sample and for each CY, a weighted average DA-level smoking rate was constructed using provincial smoking rates by education level weighted by the share of the adult population in each DA with that level of education. This composite smoking rate was included as an additional control variable in the regressions. Other control variables included indicators for the province in which the DA was located to allow for differences across provincial health systems, and the proportion of the adult population in the DA that had a university degree. In the presence of income quintile, education might reflect differences across DAs in other health behaviors important to HNC. We estimated a generalized negative binomial count model where the dependent variable was the number of cases of a particular type of HNC that occurred in a DA in the range of years associated with each CY. Since a majority of DAs in any census period have zero cases of HNC, the use of the generalized negative binomial is more appropriate than the Poisson model. For the statistical analysis, regressions were estimated separately for each of three distinct subgroups of HNC, categorized in decreasing order of possible association with human papillomavirus (HPV), as in a prior study from our group [11]: (i) cancers of the oropharynx (ICD-O-2/3 codes C019, C024, C051, C052, C090-103, C108-109, C140, and C142); (ii) cancers of the oral cavity (C0034, C020-C023, C029-C031, C039-C041, C049-50, C059-C062, and C069); and (iii) cancers of the hypopharynx, larynx, and nasopharynx (C110-C139, C300, and C310-C329).
1074
E. Hwang et al. / Oral Oncology 49 (2013) 1072–1076
Results
Other subsites
Oropharyngeal cancer
For cancers of the hypopharynx, larynx, and nasopharynx, a significant negative correlation between incidence and InQ was also apparent (Table 1). In the reference year, the IR for these cancers among people in the lowest InQ was 48% higher than in the highest InQ (IRR 1.48, p = 0.00, CI 1.36-1.62) and the IR among people in the second lowest InQ was 34% higher (IRR 1.34, p = 0.00, CI 1.23-1.46). As with cancers of the oral cavity, there was no significant trend over time in incidence rate (Table 2) for the reference group. As well, the gap in the incidence between lower and higher InQs did not narrow over time except that incidence rates in the 2006 CY were lower for all InQs relative to the highest InQ. Incidence was strongly positively correlated with smoking rate (IRR 2.54, p = 0.01, CI 1.29-4.97) (Table 3). Thus, the decline in the incidence of these types of cancers reported in our previous work is in fact reflecting declining smoking rates over time, and controlling for changes in smoking rate indicates that there is no underlying trend in incidence [10]. The highest incidence rate across provinces was in Quebec (IRR 1.32, p = 0.00, CI 1.25-1.39). As well, incidence of these cancers were lower in DAs with higher proportions of university graduates (IRR 0.50, p = 0.00, CI 0.43-0.57).
For the CY 1991, the reference year in the regression (corresponding to cancers diagnosed in the years 1992-1995), the incidence rate (IR) of oropharyngeal cancer per 100,000 person-years was significantly and negatively correlated with InQ (Table 1). The IR in the lowest InQ was 66% higher than that in the highest InQ (IRR 1.66, p = 0.00, CI 1.44-1.92). The second lowest InQ’s IR was 54% higher (IRR 1.54, p = 0.00, CI 1.34-1.77). The middle InQ’s IR was 27% higher (IRR 1.32, p = 0.00, CI 1.10-1.46), while there was no significant difference between the IRs of the second highest and highest InQs. The incidence of oropharyngeal cancer increased over the time period 1992–2007 for all InQs (Table 2). However, the greatest increase in incidence was in the highest InQs. As a result, over the time period the gap between the incidence of the highest and lowest InQs significantly narrowed, illustrated by the fact that the IRRs on lower InQs in later years were mostly less than 1 (Table 2). By the 2006 CY, the gap in incidence between highest and lowest InQs was only 75% as large as in the 1991 CY (IRR 0.75, p = 0.00, CI 0.62-0.90). The gap in incidence between highest and second lowest InQ narrowed by the same proportion (IRR 0.75, p = 0.00, CI 0.62-0.91). There was no significant effect of changes in the smoking rate on the incidence of oropharyngeal cancer (IRR 1.00, p = 0.33, CI 0.99-1.01) (Table 3). There were significant differences in incidence across provinces; the highest incidence rate was found in Nova Scotia (IRR 1.53, p = 0.00, CI 1.38-1.69). A higher proportion of adults with a university degree was associated with a lower incidence of oropharyngeal cancer (IRR 0.72, p = 0.00, CI 0.60-0.88). Oral cavity cancer For oral cavity cancer, a similar pattern for the relationship between incidence and InQ was evident although the magnitudes of the estimated IRRs were smaller than for oropharyngeal cancer (Table 1). The IR for the lowest InQ in the reference year was 48% higher than that in the highest InQ (IRR 1.48, p = 0.00, CI 1.301.67), while the IR for the second lowest InQ was 29% higher (IRR 1.29, p = 0.00, CI 1.14-1.46). However, there was no significant change in the incidence rate of oral cavity cancer over time for any of the InQs (Table 2). Similarly, there was no significant narrowing of the gap in incidence between the highest and lowest InQs over the time period. As with oropharyngeal cancer, smoking rate changes were also not a significant influence on the incidence of oral cavity cancer (IRR 0.92, p = 0.85, CI 0.38-2.25) (Table 3). There were fewer differences in incidence of oral cancer apparent across provinces, and there was no significant relationship between incidence and the proportion of the adult population with a university degree (IRR 0.95, p = 0.61, CI 0.79-1.15).
Discussion Our study confirmed our a priori hypothesis that the incidence of each of the three subsites of HNC was strongly negatively correlated with household InQ, but only oropharyngeal cancers showed a significant change in incidence over our sample period. There are two dimensions to this time pattern: first, incidence increased markedly over time, and second, this increase was most pronounced for the highest InQs. The net effect of these two dimensions was that the incidence gap between higher and lower InQs narrowed significantly over time. Our results showing a strong relationship between SES and HNC incidence across Canada using national Cancer Registry data confirmed our earlier results in Ottawa, Ontario, and in the United States [4,8]. Similar results have been identified in other countries. In the United Kingdom, Edwards and Jones conducted a study using data from four cancer registries from 1984 to 1993 and also noted that people living in deprived areas were more likely to develop upper aerodigestive tract cancer than people living in wealthy areas [11]. In Denmark, Andersen et al. evaluated how socioeconomic, demographic, and health-related indicators influenced the incidence of mouth, pharynx, and larynx cancers diagnosed from 1994 to 2003 using data from nationwide registries. Likewise, the incidence of all three types of HNC increased with decreasing socioeconomic position, measured by disposable income, work market affiliation, social class, housing tenure, cohabiting status, and type of district [12].
Table 1 Incidence of head and neck cancer in census year 1991 (reference year) by income quintile. Oropharynx IRR Q5 (reference quintile) Q4 Q3 Q2 Q1
1.00 1.08 1.27* 1.54* 1.66*
Oral cavity p 0.31 0.00 0.00 0.00
95% CI
IRR
0.93–1.25 1.10–1.46 1.34–1.77 1.44–1.92
1.00 1.03 1.07 1.29* 1.48*
Hypopharynx, larynx, nasopharynx p 0.61 0.30 0.00 0.00
95% CI
IRR
p
95% CI
0.91–1.17 0.94–1.21 1.14–1.46 1.31–1.67
1.00 1.08 1.21* 1.34* 1.48*
0.11 0.00 0.00 0.00
0.98–1.18 1.11–1.32 1.23–1.46 1.36–1.62
Q5, highest income quintile; Q4, second highest income quintile; Q3, middle income quintile; Q2, second lowest income quintile; Q1, lowest income quintile; IRR, incidence rate ratio; p, p value; CI, confidence interval. * Statistically significant.
1075
E. Hwang et al. / Oral Oncology 49 (2013) 1072–1076 Table 2 Incidence of head and neck cancer over time. Oropharynx IRR CY 1991, Q5 (reference year and InQ) CY 1996 CY 2001 CY 2006 CY1996xQ4 CY2001xQ4 CY2006xQ4 CY1996xQ3 CY2001xQ3 CY2006xQ3 CY1996xQ2 CY2001xQ2 CY2006xQ2 CY1996xQ1 CY2001xQ1 CY2006xQ1
1.00 1.14* 1.43* 1.57* 0.97 0.97 0.99 0.87 0.90 0.82* 0.91 0.78* 0.75* 0.95 0.87 0.75*
p 0.05 0.00 0.00 0.75 0.71 0.91 0.14 0.24 0.05 0.27 0.00 0.00 0.58 0.11 0.00
Oral cavity 95% CI
IRR
1.00–1.31 1.23–1.66 1.34–1.85 0.81–1.17 0.81–1.15 0.81–1.20 0.73–1.05 0.76–1.07 0.68–1.00 0.76–1.08 0.66–0.92 0.62–0.91 0.80–1.13 0.74–1.03 0.62–0.90
1.00 0.99 1.03 0.98 1.05 1.00 1.05 1.03 1.04 1.04 0.95 0.88 1.02 0.97 0.88 0.84
Hypopharynx, larynx, nasopharynx p 0.82 0.72 0.77 0.57 0.99 0.64 0.7 0.58 0.69 0.52 0.10 0.82 0.69 0.11 0.06
95% CI
IRR
p
95% CI
0.88–1.11 0.90–1.17 0.84–1.14 0.89–1.23 0.86–1.17 0.87–1.26 0.88–1.21 0.89–1.22 0.86–1.20 0.81–1.11 0.75–1.03 0.86–1.26 0.83–1.13 0.76–1.03 0.71–1.00
1.00 1.01 1.11* 1.04 0.99 1.04 0.90 0.91 0.93 0.87* 0.94 0.93 0.86* 0.99 0.96 0.86*
0.77 0.05 0.47 0.81 0.56 0.13 0.11 0.23 0.05 0.25 0.21 0.03 0.91 0.45 0.03
0.93–1.11 1.00–1.22 0.93–1.17 0.88–1.11 0.92–1.16 0.78–1.03 0.81–1.02 0.83–1.05 0.75–1.00 0.83–1.05 0.83–1.04 0.75–0.99 0.89–1.11 0.85–1.07 0.75–0.98
CY, census year; Q4, second highest income quintile; Q3, middle income quintile; Q2, second lowest income quintile; Q1, lowest income quintile; IRR, incidence rate ratio; p, p value; CI, confidence interval. * Statistically significant.
Table 3 Other determinants of the incidence of head and neck cancer. Oropharynx
Smoking rate University degree Ontario Newfoundland PEI Nova Scotia New Brunswick Quebec Manitoba Saskatchewan Alberta BC
Oral cavity
Hypopharynx, larynx, nasopharynx
IRR
p
95% CI
IRR
p
95% CI
IRR
p
95% CI
0.64 0.72* 1.00 0.99 1.34* 1.53* 1.25* 1.23* 1.19* 1.00 1.01 1.23*
0.35 0.00
0.25–1.62 0.60–0.88
0.85 0.61
0.38–2.25 0.79–1.15
1.29–4.97 0.43–0.57
0.85–1.15 1.05–1.71 1.38–1.69 1.11–1.40 1.14–1.32 1.09–1.31 0.89–1.12 0.94–1.09 1.16–1.31
0.01 0.45 0.02 0.64 0.00 0.01 0.29 0.08 0.00
0.70–0.95 0.86–1.40 1.03–1.27 0.86–1.09 0.76–0.89 1.03–1.23 0.84–1.05 0.99–1.13 1.06–1.19
2.54* 0.50* 1.00 1.09 1.19 1.06 1.22* 1.32* 1.00 0.91* 0.91* 1.12*
0.01 0.00
0.90 0.02 0.00 0.00 0.00 0.00 0.98 0.71 0.00
0.92 0.95 1.00 0.81* 1.10 1.14* 0.97 0.82* 1.12* 0.94 1.06 1.12*
0.10 0.06 0.15 0.00 0.00 0.97 0.04 0.00 0.00
0.98–1.20 0.99–1.42 0.98–1.15 1.13–1.33 1.25–1.39 0.93–1.08 0.84–0.99 0.86–0.96 1.07–1.17
IRR, incidence rate ratio; p, p value; CI, confidence interval; Smoking Rate is an education-specific composite provincial smoking rate; University Degree is the proportion of adult residents in the DA that have completed a university degree. * Statistically significant.
Overall IRs of oropharyngeal cancer, which has a high association with HPV, increased over our study period. This growth in incidence is in keeping with the epidemiology of HPV-related oropharyngeal cancer explained in recent studies [13,14]. The narrowing of the gap in incidence between the lowest and highest InQs over time in the context of overall increase in incidence indicates that the incidence of oropharyngeal cancer in the highest InQs increased relatively faster than in the lowest InQs. We questioned whether this phenomenon was due to a concurrent narrowing in the income gap between the lowest and highest InQs. We confirmed that the opposite in fact occurred as the income gap widened over the period of our study. The ratio of real (inflationadjusted) average household income in the highest InQ relative to the lowest InQ was 2.42 in CY 1991, 2.53 in CY 1996, 2.71 in CY 2001, and 2.68 in CY 2006. The rapid rise in incidence seen with higher income raises the possibility that this trend might be associated with the changes in the epidemiology of oropharyngeal cancer which is now mostly attributable to HPV infection and diagnosed in younger individuals [14] who could potentially tend to be more educated and affluent. We are limited in the fact that HPV data was unavailable for patients recorded in the CCR during the study period. Although further studies are required to support this implication, we can at least make the recommendation, based on this finding, that clinicians who treat HNC be aware of the
increasing incidence of oropharyngeal cancer in wealthier individuals. For oral cavity cancer and cancers of the hypopharynx, larynx, and nasopharynx, the overall IRs remained stable over the four CYs in our study after controlling for smoking rate and demographic factors. Also, there was no significant time trend in the difference in incidence between the lowest and highest InQs as was seen in oropharyngeal cancer. For cancers of the hypopharynx, larynx, and nasopharynx, the overall IRs fell over time once smoking controls were removed, reflecting the effect of reduction in smoking rate over time. In addition to the aforementioned lack of HPV status information, there are two other limitations of our study. First, we did not possess individual-level income data and thus used DA-level information on household income as a substitute, which is less precise than individual-level information. Furthermore, while household income reflects income from all sources including pensions and investment income, socioeconomic status could be measured more accurately with individual-level information on other determinants of socioeconomic status such as education level and occupation. Income alone may be biased in favor of current socioeconomic status rather than long run socioeconomic status. Second, as data on prevalence of smoking at the DA-level was unavailable, we used provincial education-specific smoking rates
1076
E. Hwang et al. / Oral Oncology 49 (2013) 1072–1076
and DA-level education distribution to construct a composite smoking rate. In conclusion, in Canada, from 1992 to 2007, the incidence of HNC was higher in lower income groups for all HNC subsites after controlling for age, gender, province, time, and smoking. The only subsite that had a significant increase in incidence over the period of the study was oropharyngeal cancer. Oropharyngeal cancer also showed a significant narrowing of the gap in incidence between the lowest and highest InQs. To our knowledge, this is the first time nationwide Cancer Registry data have been available and used to study the relationship between HNC incidence and SES, defined by income, in Canada. Our findings confirm the inverse relationship between HNC incidence and income identified in the US and Europe despite the Canadian universal health care system. Further studies are warranted to understand the mechanisms behind this socioeconomic discrepancy. Conflict of Interest None declared. References [1] Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA Cancer J Clin 2005;55:74–108.
[2] Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J Clin 2012;62:10–29. [3] MacKillop WJ, Zhang-Salomons J, Boyd CJ, Groome PA. Associations between community income and cancer incidence in Canada and the United States. Cancer 2000;89:901–12. [4] Johnson S, McDonald JT, Corsten M. Socioeconomic factors in head and neck cancer. J Otolaryngol Head Neck Surg 2008;37:597–601. [5] Conway DI, Brewster DH, McKinney PA, Stark J, McMahon AD, Macpherson LMD. Widening socio-economic inequalities in oral cancer incidence in Scotland, 1976–2002. Br J Cancer 2007;96:818–20. [6] Greenwood M, Thomson PJ, Lowry RJ, Steen IN. Oral cancer: material deprivation, unemployment and risk factor behaviour – an initial study. Int J Oral Maxillofac Surg 2003;32:74–7. [7] Greenberg RS, Haber MJ, Clark WS, et al. The relation of socioeconomic status to oral and pharyngeal cancer. Epidemiology 1991;2:194–200. [8] Johnson S, McDonald JT, Corsten M, Rourke R. Socio-economic status and head and neck cancer incidence in Canada: a case-control study. Oral Oncol 2010;46:200–3. [9] Arsenijevic S, Pantovic V, Gledovic Z, Stojanovic J, Belic B. Demographic characteristics of patients with laryngeal cancer and their socioeconomic status. J BUON 2010;15:131–5. [10] Johnson-Obaseki S, McDonald JT, Corsten M, Rourke R. Head and neck cancer in Canada: trends 1992 to 2007. Otolaryngol Head Neck Surg 2012;147:74–8. [11] Edwards DM, Jones J. Incidence of and survival from upper aerodigestive tract cancers in the UK: the influence of deprivation. Eur J Cancer 1999;35:968–72. [12] Andersen ZJ, Lassen CF, Clemmesen IH. Social inequality and incidence of and survival from cancers of the mouth, pharynx and larynx in a population-based study in Denmark, 1994–2003. Eur J Cancer 2008;44:1950–61. [13] Simard EP, Ward EM, Siegel R, Jamal A. Cancers with increasing incidence trends in the United States: 1999 through 2008. CA Cancer J Clin 2012;62:118–28. [14] Chaturvedi AK. Epidemiology and clinical aspects of HPV in head and neck cancers. Head Neck Pathol 2012;6:S16–24.
Contents lists available at ScienceDirect
Oral Oncology journal homepage: www.elsevier.com/locate/oraloncology
Incidence of head and neck cancer and socioeconomic status in Canada from 1992 to 2007 Euna Hwang a, Stephanie Johnson-Obaseki a, James Ted McDonald b, Chris Connell b, Martin Corsten a,⇑ a b
Department of Otolaryngology-Head and Neck Surgery, University of Ottawa, Ottawa, Canada Department of Economics, University of New Brunswick, Fredericton, Canada
a r t i c l e
i n f o
Article history: Received 25 January 2013 Received in revised form 5 May 2013 Accepted 14 August 2013 Available online 7 September 2013 Keywords: Head and neck cancer Incidence Socioeconomic status Income Oropharynx cancer Human papillomavirus
s u m m a r y Objective: To investigate the association between head and neck cancer (HNC) incidence and socioeconomic status (SES) in Canada, and investigate changes in the relationship between HNC incidence and SES from 1992 to 2007. Materials and methods: Cases were drawn from the Canadian Cancer Registry (1992–2007), and were categorized into three subsites: oropharynx, oral cavity, and ‘‘other’’ (hypopharynx, larynx, and nasopharynx). Demographic and socioeconomic information were extracted from the Canadian Census of Population data for the study period (1992–2007), which included four census years. We linked cases to income quintiles (InQs) according to patients’ postal codes. Results: Incidence rates in the lowest InQs were significantly higher than in the highest InQs for all HNC subsites. The incidence of oropharyngeal cancer increased over the time period 1992–2007 for all InQs. However, the greatest increase in incidence was in the highest InQs. As a result, over the time period the gap between the incidence of the highest and lowest InQs significantly narrowed for oropharyngeal cancer. For oral cavity cancer and the other head and neck cancers, the overall incidence did not increase and the gap in incidence did not change significantly. Conclusion: HNC incidence was higher with lower income in all HNC subsites in Canada from 1992 to 2007. The gap in incidence between the highest and the lowest InQs narrowed for oropharyngeal cancer over the time period studied, but was unchanged for the other HNC subsites. Ó 2013 Elsevier Ltd. All rights reserved.
Introduction Head and neck cancer (HNC) accounts for more than 550,000 cases annually worldwide and represents approximately 3% of malignancies in the United States (US) [1,2]. Multiple studies from the US, Canada, and Europe have revealed that HNC is more frequent in socioeconomically-deprived populations [4–9]. Although this observation is mostly attributed to confounding risk factors associated with lower socioeconomic status (SES) such as tobacco and alcohol consumption and poor oral hygiene, our group and others have shown that there is a higher incidence of HNC with lower SES after controlling for such factors [4,7,8]. The underlying etiology behind this association remains unclear and is likely multifactorial. Moreover, there is no consensus in the literature for parameters of SES when studying the relationship between SES and the epidemiology of cancer.
⇑ Corresponding author. Address: Department of Otolaryngology-Head and Neck Surgery, University of Ottawa/The Ottawa Hospital, Suite S-3, 501 Smyth Road, Ottawa, ON K1H 8L6, Canada. Tel.: +1 (613) 737 8596; fax: +1 (613) 739 6542. E-mail address: [email protected] (M. Corsten). 1368-8375/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.oraloncology.2013.08.002
Our group previously examined the association between SES and HNC incidence in both the US (National Institutes of Health Survey) and Canada (case-control study conducted in the province of Ontario) by looking at various measures of SES. In the US, the following factors were associated with a higher incidence of HNC: marital status of single, never married; lower education (maximum level of education achieved less than high school completion); as well as lower income (family income of less than US$20,000) [4]. In Ontario, Canada, an increased incidence of HNC was similarly associated with lower education (less than grade 8 education) and lower median family income [8]. When specifically assessing income, MacKillop et al. concluded that there were strong, inverse associations between income level and the incidence of cancers of the head and neck region in Ontario and the US [3]. In addition to the effects of SES on HNC incidence, there have been dramatic changes in the incidence of cancer at some head and neck subsites, particularly oropharynx, over the last two decades. This change in incidence is related to the increase in the number of head and neck cancers caused by human papilloma virus over the same time period. On the other hand, the incidence of cancer at other head and neck subsites, such as larynx, has been
E. Hwang et al. / Oral Oncology 49 (2013) 1072–1076
decreasing over that time period, largely due to the decrease in smoking incidence over that time. The incidence of oral cavity cancer has been relatively stable. Our group sought to investigate the in the SES/HNC incidence relationship over the time period 1992– 2007. Our a priori hypothesis was that the magnitude of the relationship between SES and HNC incidence for oropharynx cancer would decrease over the time period, whereas the magnitude of the relationship between SES and HNC incidence for other head and neck cancers not predominantly caused by HPV would remain stable. To our knowledge, there has not been a Canadian study using nationwide Cancer Registry data to examine the relationship between HNC incidence and SES. The goals of the study were twofold: (1) to confirm the relationship between SES and HNC incidence using Canadian national Registry data and; (2) to investigate the changes in the relationship between SES and HNC incidence over time.
Patients and methods Our data were drawn from two sources: the Canadian Cancer Registry (CCR) data file and the Canadian Census of Population from Statistics Canada. The CCR data file contains patient demographic and tumor-specific data on each tumor recorded in provincial and territorial cancer registries from 1992 to 2007, inclusive. Names and personal identification numbers were removed and replaced with unique identifiers in the version of the file released for this study. All analysis was conducted within the New Brunswick Research Data Centre (NB-RDC) and all output was vetted for release using enhanced vetting methods required by Statistics Canada. Our study complied with the University of New Brunswick Institutional Review Board ethics requirements, which do not necessitate an additional review for research projects using Statistics Canada data stored in the NB-RDC. To our knowledge, this is the first time CCR data have been available and used to study the relationship between HNC incidence and SES in Canada. The Canadian Census of Population contains demographic and socioeconomic information from each census, which occurs every five years in Canada. Data were available for the census years (CYs) 1991, 1996, 2001, and 2006. The finest level of disaggregation at which census information is released by Statistics Canada is the Dissemination Area (DA), a small, stable geographic unit composed of one or more adjacent blocks, with a population of 400–700 persons. DAs cover the whole territory of Canada. For each DA and for each of the four CYs, information was collected on age for men and women, average household income, and proportion of adult residents with at least a university degree. Although average income in the DA might be influenced by large income outliers, median income is not available at the DA-level for all of the CYs. The number of DAs for which census information was available was 32,825 in 1991, 38,016 in 1996, 46,909 in 2001, and 52,443 in 2006. In order to control for wage and price differences across different regions, the income quintile (InQ) for each DA was defined relative to other DAs in the associated Census Division (CD), which Statistics Canada defines as a group of neighboring municipalities joined together for the purposes of regional planning and managing common services. The number of CDs in Canada was in the range of 288–290 across the four CYs and changes occurred to only a few CD boundaries each CY. DAs within each CD were sorted by average income and then assigned to one of five InQs. DAs in a CD are ranked by average household income from highest to lowest. Each DA is assigned to an InQ based on its rank in the distribution. Measuring with respect to the CD allows for variation in prices and purchasing power across Canadian regions. Although the DA of an individual in the CCR database
1073
is not reported, the postal code of residence is disclosed. PCCF+ is a statistical tool provided by Statistics Canada that maps postal codes to DAs. Statistics Canada revises geographic boundaries and updates PCCF+ at each CY. Hence, in order to match postal code and DA information as closely as possible, HNC cases were assigned to CYs as follows: cases diagnosed in 1992–1995 were associated with socioeconomic characteristics according to the 1991 Census; cases diagnosed in 1996–2000 were associated with data from the 1996 Census; cases diagnosed in 2001–2005 were associated with data from the 2001 Census; and cases diagnosed in 2006–2007 were associated with data from the 2006 Census. Census data could be linked to 96.9% of DAs for the CY 1991, 97.4% for the CY 1996, 99.0% for the CY 2001, and 99.2% for the CY 2006. For each CY, the unit of observation for the analysis of incidence was the DA. The key variable of interest was the number of cases of different types of HNC diagnosed in adults over the age of 18 in each DA over the relevant period of time corresponding to that CY. At the DA-level, it is not feasible to disaggregate counts further by age and gender. In the regression analysis, controls were included for the age and gender composition of the adult population of each DA. The exposure variable was the adult population in the DA at the CY multiplied by the number of years in the corresponding time interval for that census (2, 4, or 5 years). Poisson regression models were estimated where neighborhood SES as measured by the InQ of the DA was captured by a 0/1 binary variable for each InQ, with the highest InQ specified as the baseline. The regressions also included indicator variables for each province or territory of residence in order to capture regional unobserved effects important to the incidence and diagnosis of HNC. To capture changes in incidence over time, indicator variables were defined for each CY from 1996 to 2006 with 1991 as the reference year. These CY variables interacted with each InQ to allow the incidence of HNC to vary by average household income over time. One of the main risk factors for most forms of HNC is smoking; however, data on smoking prevalence was not available at the DAlevel from Statistics Canada surveys. As a substitute, provincial smoking rates by education level were obtained for each CY from various waves of the Canadian Community Health Survey and the National Population Health Survey. For each DA in our sample and for each CY, a weighted average DA-level smoking rate was constructed using provincial smoking rates by education level weighted by the share of the adult population in each DA with that level of education. This composite smoking rate was included as an additional control variable in the regressions. Other control variables included indicators for the province in which the DA was located to allow for differences across provincial health systems, and the proportion of the adult population in the DA that had a university degree. In the presence of income quintile, education might reflect differences across DAs in other health behaviors important to HNC. We estimated a generalized negative binomial count model where the dependent variable was the number of cases of a particular type of HNC that occurred in a DA in the range of years associated with each CY. Since a majority of DAs in any census period have zero cases of HNC, the use of the generalized negative binomial is more appropriate than the Poisson model. For the statistical analysis, regressions were estimated separately for each of three distinct subgroups of HNC, categorized in decreasing order of possible association with human papillomavirus (HPV), as in a prior study from our group [11]: (i) cancers of the oropharynx (ICD-O-2/3 codes C019, C024, C051, C052, C090-103, C108-109, C140, and C142); (ii) cancers of the oral cavity (C0034, C020-C023, C029-C031, C039-C041, C049-50, C059-C062, and C069); and (iii) cancers of the hypopharynx, larynx, and nasopharynx (C110-C139, C300, and C310-C329).
1074
E. Hwang et al. / Oral Oncology 49 (2013) 1072–1076
Results
Other subsites
Oropharyngeal cancer
For cancers of the hypopharynx, larynx, and nasopharynx, a significant negative correlation between incidence and InQ was also apparent (Table 1). In the reference year, the IR for these cancers among people in the lowest InQ was 48% higher than in the highest InQ (IRR 1.48, p = 0.00, CI 1.36-1.62) and the IR among people in the second lowest InQ was 34% higher (IRR 1.34, p = 0.00, CI 1.23-1.46). As with cancers of the oral cavity, there was no significant trend over time in incidence rate (Table 2) for the reference group. As well, the gap in the incidence between lower and higher InQs did not narrow over time except that incidence rates in the 2006 CY were lower for all InQs relative to the highest InQ. Incidence was strongly positively correlated with smoking rate (IRR 2.54, p = 0.01, CI 1.29-4.97) (Table 3). Thus, the decline in the incidence of these types of cancers reported in our previous work is in fact reflecting declining smoking rates over time, and controlling for changes in smoking rate indicates that there is no underlying trend in incidence [10]. The highest incidence rate across provinces was in Quebec (IRR 1.32, p = 0.00, CI 1.25-1.39). As well, incidence of these cancers were lower in DAs with higher proportions of university graduates (IRR 0.50, p = 0.00, CI 0.43-0.57).
For the CY 1991, the reference year in the regression (corresponding to cancers diagnosed in the years 1992-1995), the incidence rate (IR) of oropharyngeal cancer per 100,000 person-years was significantly and negatively correlated with InQ (Table 1). The IR in the lowest InQ was 66% higher than that in the highest InQ (IRR 1.66, p = 0.00, CI 1.44-1.92). The second lowest InQ’s IR was 54% higher (IRR 1.54, p = 0.00, CI 1.34-1.77). The middle InQ’s IR was 27% higher (IRR 1.32, p = 0.00, CI 1.10-1.46), while there was no significant difference between the IRs of the second highest and highest InQs. The incidence of oropharyngeal cancer increased over the time period 1992–2007 for all InQs (Table 2). However, the greatest increase in incidence was in the highest InQs. As a result, over the time period the gap between the incidence of the highest and lowest InQs significantly narrowed, illustrated by the fact that the IRRs on lower InQs in later years were mostly less than 1 (Table 2). By the 2006 CY, the gap in incidence between highest and lowest InQs was only 75% as large as in the 1991 CY (IRR 0.75, p = 0.00, CI 0.62-0.90). The gap in incidence between highest and second lowest InQ narrowed by the same proportion (IRR 0.75, p = 0.00, CI 0.62-0.91). There was no significant effect of changes in the smoking rate on the incidence of oropharyngeal cancer (IRR 1.00, p = 0.33, CI 0.99-1.01) (Table 3). There were significant differences in incidence across provinces; the highest incidence rate was found in Nova Scotia (IRR 1.53, p = 0.00, CI 1.38-1.69). A higher proportion of adults with a university degree was associated with a lower incidence of oropharyngeal cancer (IRR 0.72, p = 0.00, CI 0.60-0.88). Oral cavity cancer For oral cavity cancer, a similar pattern for the relationship between incidence and InQ was evident although the magnitudes of the estimated IRRs were smaller than for oropharyngeal cancer (Table 1). The IR for the lowest InQ in the reference year was 48% higher than that in the highest InQ (IRR 1.48, p = 0.00, CI 1.301.67), while the IR for the second lowest InQ was 29% higher (IRR 1.29, p = 0.00, CI 1.14-1.46). However, there was no significant change in the incidence rate of oral cavity cancer over time for any of the InQs (Table 2). Similarly, there was no significant narrowing of the gap in incidence between the highest and lowest InQs over the time period. As with oropharyngeal cancer, smoking rate changes were also not a significant influence on the incidence of oral cavity cancer (IRR 0.92, p = 0.85, CI 0.38-2.25) (Table 3). There were fewer differences in incidence of oral cancer apparent across provinces, and there was no significant relationship between incidence and the proportion of the adult population with a university degree (IRR 0.95, p = 0.61, CI 0.79-1.15).
Discussion Our study confirmed our a priori hypothesis that the incidence of each of the three subsites of HNC was strongly negatively correlated with household InQ, but only oropharyngeal cancers showed a significant change in incidence over our sample period. There are two dimensions to this time pattern: first, incidence increased markedly over time, and second, this increase was most pronounced for the highest InQs. The net effect of these two dimensions was that the incidence gap between higher and lower InQs narrowed significantly over time. Our results showing a strong relationship between SES and HNC incidence across Canada using national Cancer Registry data confirmed our earlier results in Ottawa, Ontario, and in the United States [4,8]. Similar results have been identified in other countries. In the United Kingdom, Edwards and Jones conducted a study using data from four cancer registries from 1984 to 1993 and also noted that people living in deprived areas were more likely to develop upper aerodigestive tract cancer than people living in wealthy areas [11]. In Denmark, Andersen et al. evaluated how socioeconomic, demographic, and health-related indicators influenced the incidence of mouth, pharynx, and larynx cancers diagnosed from 1994 to 2003 using data from nationwide registries. Likewise, the incidence of all three types of HNC increased with decreasing socioeconomic position, measured by disposable income, work market affiliation, social class, housing tenure, cohabiting status, and type of district [12].
Table 1 Incidence of head and neck cancer in census year 1991 (reference year) by income quintile. Oropharynx IRR Q5 (reference quintile) Q4 Q3 Q2 Q1
1.00 1.08 1.27* 1.54* 1.66*
Oral cavity p 0.31 0.00 0.00 0.00
95% CI
IRR
0.93–1.25 1.10–1.46 1.34–1.77 1.44–1.92
1.00 1.03 1.07 1.29* 1.48*
Hypopharynx, larynx, nasopharynx p 0.61 0.30 0.00 0.00
95% CI
IRR
p
95% CI
0.91–1.17 0.94–1.21 1.14–1.46 1.31–1.67
1.00 1.08 1.21* 1.34* 1.48*
0.11 0.00 0.00 0.00
0.98–1.18 1.11–1.32 1.23–1.46 1.36–1.62
Q5, highest income quintile; Q4, second highest income quintile; Q3, middle income quintile; Q2, second lowest income quintile; Q1, lowest income quintile; IRR, incidence rate ratio; p, p value; CI, confidence interval. * Statistically significant.
1075
E. Hwang et al. / Oral Oncology 49 (2013) 1072–1076 Table 2 Incidence of head and neck cancer over time. Oropharynx IRR CY 1991, Q5 (reference year and InQ) CY 1996 CY 2001 CY 2006 CY1996xQ4 CY2001xQ4 CY2006xQ4 CY1996xQ3 CY2001xQ3 CY2006xQ3 CY1996xQ2 CY2001xQ2 CY2006xQ2 CY1996xQ1 CY2001xQ1 CY2006xQ1
1.00 1.14* 1.43* 1.57* 0.97 0.97 0.99 0.87 0.90 0.82* 0.91 0.78* 0.75* 0.95 0.87 0.75*
p 0.05 0.00 0.00 0.75 0.71 0.91 0.14 0.24 0.05 0.27 0.00 0.00 0.58 0.11 0.00
Oral cavity 95% CI
IRR
1.00–1.31 1.23–1.66 1.34–1.85 0.81–1.17 0.81–1.15 0.81–1.20 0.73–1.05 0.76–1.07 0.68–1.00 0.76–1.08 0.66–0.92 0.62–0.91 0.80–1.13 0.74–1.03 0.62–0.90
1.00 0.99 1.03 0.98 1.05 1.00 1.05 1.03 1.04 1.04 0.95 0.88 1.02 0.97 0.88 0.84
Hypopharynx, larynx, nasopharynx p 0.82 0.72 0.77 0.57 0.99 0.64 0.7 0.58 0.69 0.52 0.10 0.82 0.69 0.11 0.06
95% CI
IRR
p
95% CI
0.88–1.11 0.90–1.17 0.84–1.14 0.89–1.23 0.86–1.17 0.87–1.26 0.88–1.21 0.89–1.22 0.86–1.20 0.81–1.11 0.75–1.03 0.86–1.26 0.83–1.13 0.76–1.03 0.71–1.00
1.00 1.01 1.11* 1.04 0.99 1.04 0.90 0.91 0.93 0.87* 0.94 0.93 0.86* 0.99 0.96 0.86*
0.77 0.05 0.47 0.81 0.56 0.13 0.11 0.23 0.05 0.25 0.21 0.03 0.91 0.45 0.03
0.93–1.11 1.00–1.22 0.93–1.17 0.88–1.11 0.92–1.16 0.78–1.03 0.81–1.02 0.83–1.05 0.75–1.00 0.83–1.05 0.83–1.04 0.75–0.99 0.89–1.11 0.85–1.07 0.75–0.98
CY, census year; Q4, second highest income quintile; Q3, middle income quintile; Q2, second lowest income quintile; Q1, lowest income quintile; IRR, incidence rate ratio; p, p value; CI, confidence interval. * Statistically significant.
Table 3 Other determinants of the incidence of head and neck cancer. Oropharynx
Smoking rate University degree Ontario Newfoundland PEI Nova Scotia New Brunswick Quebec Manitoba Saskatchewan Alberta BC
Oral cavity
Hypopharynx, larynx, nasopharynx
IRR
p
95% CI
IRR
p
95% CI
IRR
p
95% CI
0.64 0.72* 1.00 0.99 1.34* 1.53* 1.25* 1.23* 1.19* 1.00 1.01 1.23*
0.35 0.00
0.25–1.62 0.60–0.88
0.85 0.61
0.38–2.25 0.79–1.15
1.29–4.97 0.43–0.57
0.85–1.15 1.05–1.71 1.38–1.69 1.11–1.40 1.14–1.32 1.09–1.31 0.89–1.12 0.94–1.09 1.16–1.31
0.01 0.45 0.02 0.64 0.00 0.01 0.29 0.08 0.00
0.70–0.95 0.86–1.40 1.03–1.27 0.86–1.09 0.76–0.89 1.03–1.23 0.84–1.05 0.99–1.13 1.06–1.19
2.54* 0.50* 1.00 1.09 1.19 1.06 1.22* 1.32* 1.00 0.91* 0.91* 1.12*
0.01 0.00
0.90 0.02 0.00 0.00 0.00 0.00 0.98 0.71 0.00
0.92 0.95 1.00 0.81* 1.10 1.14* 0.97 0.82* 1.12* 0.94 1.06 1.12*
0.10 0.06 0.15 0.00 0.00 0.97 0.04 0.00 0.00
0.98–1.20 0.99–1.42 0.98–1.15 1.13–1.33 1.25–1.39 0.93–1.08 0.84–0.99 0.86–0.96 1.07–1.17
IRR, incidence rate ratio; p, p value; CI, confidence interval; Smoking Rate is an education-specific composite provincial smoking rate; University Degree is the proportion of adult residents in the DA that have completed a university degree. * Statistically significant.
Overall IRs of oropharyngeal cancer, which has a high association with HPV, increased over our study period. This growth in incidence is in keeping with the epidemiology of HPV-related oropharyngeal cancer explained in recent studies [13,14]. The narrowing of the gap in incidence between the lowest and highest InQs over time in the context of overall increase in incidence indicates that the incidence of oropharyngeal cancer in the highest InQs increased relatively faster than in the lowest InQs. We questioned whether this phenomenon was due to a concurrent narrowing in the income gap between the lowest and highest InQs. We confirmed that the opposite in fact occurred as the income gap widened over the period of our study. The ratio of real (inflationadjusted) average household income in the highest InQ relative to the lowest InQ was 2.42 in CY 1991, 2.53 in CY 1996, 2.71 in CY 2001, and 2.68 in CY 2006. The rapid rise in incidence seen with higher income raises the possibility that this trend might be associated with the changes in the epidemiology of oropharyngeal cancer which is now mostly attributable to HPV infection and diagnosed in younger individuals [14] who could potentially tend to be more educated and affluent. We are limited in the fact that HPV data was unavailable for patients recorded in the CCR during the study period. Although further studies are required to support this implication, we can at least make the recommendation, based on this finding, that clinicians who treat HNC be aware of the
increasing incidence of oropharyngeal cancer in wealthier individuals. For oral cavity cancer and cancers of the hypopharynx, larynx, and nasopharynx, the overall IRs remained stable over the four CYs in our study after controlling for smoking rate and demographic factors. Also, there was no significant time trend in the difference in incidence between the lowest and highest InQs as was seen in oropharyngeal cancer. For cancers of the hypopharynx, larynx, and nasopharynx, the overall IRs fell over time once smoking controls were removed, reflecting the effect of reduction in smoking rate over time. In addition to the aforementioned lack of HPV status information, there are two other limitations of our study. First, we did not possess individual-level income data and thus used DA-level information on household income as a substitute, which is less precise than individual-level information. Furthermore, while household income reflects income from all sources including pensions and investment income, socioeconomic status could be measured more accurately with individual-level information on other determinants of socioeconomic status such as education level and occupation. Income alone may be biased in favor of current socioeconomic status rather than long run socioeconomic status. Second, as data on prevalence of smoking at the DA-level was unavailable, we used provincial education-specific smoking rates
1076
E. Hwang et al. / Oral Oncology 49 (2013) 1072–1076
and DA-level education distribution to construct a composite smoking rate. In conclusion, in Canada, from 1992 to 2007, the incidence of HNC was higher in lower income groups for all HNC subsites after controlling for age, gender, province, time, and smoking. The only subsite that had a significant increase in incidence over the period of the study was oropharyngeal cancer. Oropharyngeal cancer also showed a significant narrowing of the gap in incidence between the lowest and highest InQs. To our knowledge, this is the first time nationwide Cancer Registry data have been available and used to study the relationship between HNC incidence and SES, defined by income, in Canada. Our findings confirm the inverse relationship between HNC incidence and income identified in the US and Europe despite the Canadian universal health care system. Further studies are warranted to understand the mechanisms behind this socioeconomic discrepancy. Conflict of Interest None declared. References [1] Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA Cancer J Clin 2005;55:74–108.
[2] Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J Clin 2012;62:10–29. [3] MacKillop WJ, Zhang-Salomons J, Boyd CJ, Groome PA. Associations between community income and cancer incidence in Canada and the United States. Cancer 2000;89:901–12. [4] Johnson S, McDonald JT, Corsten M. Socioeconomic factors in head and neck cancer. J Otolaryngol Head Neck Surg 2008;37:597–601. [5] Conway DI, Brewster DH, McKinney PA, Stark J, McMahon AD, Macpherson LMD. Widening socio-economic inequalities in oral cancer incidence in Scotland, 1976–2002. Br J Cancer 2007;96:818–20. [6] Greenwood M, Thomson PJ, Lowry RJ, Steen IN. Oral cancer: material deprivation, unemployment and risk factor behaviour – an initial study. Int J Oral Maxillofac Surg 2003;32:74–7. [7] Greenberg RS, Haber MJ, Clark WS, et al. The relation of socioeconomic status to oral and pharyngeal cancer. Epidemiology 1991;2:194–200. [8] Johnson S, McDonald JT, Corsten M, Rourke R. Socio-economic status and head and neck cancer incidence in Canada: a case-control study. Oral Oncol 2010;46:200–3. [9] Arsenijevic S, Pantovic V, Gledovic Z, Stojanovic J, Belic B. Demographic characteristics of patients with laryngeal cancer and their socioeconomic status. J BUON 2010;15:131–5. [10] Johnson-Obaseki S, McDonald JT, Corsten M, Rourke R. Head and neck cancer in Canada: trends 1992 to 2007. Otolaryngol Head Neck Surg 2012;147:74–8. [11] Edwards DM, Jones J. Incidence of and survival from upper aerodigestive tract cancers in the UK: the influence of deprivation. Eur J Cancer 1999;35:968–72. [12] Andersen ZJ, Lassen CF, Clemmesen IH. Social inequality and incidence of and survival from cancers of the mouth, pharynx and larynx in a population-based study in Denmark, 1994–2003. Eur J Cancer 2008;44:1950–61. [13] Simard EP, Ward EM, Siegel R, Jamal A. Cancers with increasing incidence trends in the United States: 1999 through 2008. CA Cancer J Clin 2012;62:118–28. [14] Chaturvedi AK. Epidemiology and clinical aspects of HPV in head and neck cancers. Head Neck Pathol 2012;6:S16–24.